-Take faces. People are [verifiably very good at recognizing sex from (hair covered, males clean-shaven) photographs of people's faces](/papers/bruce_et_al-sex_discrimination_how_do_we_tell.pdf) (96% accuracy, which is the equivalent of _d_ ≈ 3.5), but we don't have direct introspective access into what _specific_ features our brains are using to do it; we just look, and _somehow_ know. The differences are real, but it's not a matter of any single, simple measurement you could perform with a ruler (like the distance between someone's eyes). Rather, it's a high-dimensional _pattern_ in many measurements you could take with a rule, no one of which is definitive. [Covering up the nose makes people slower and slightly worse at sexing faces, but people don't do better than chance at guessing sex from photos of noses alone](/papers/roberts-bruce-feature_saliency_in_judging_the_sex_and_familiarity_of_faces.pdf).
+[TODO (somewhere around-ish this section): chromosomes at the root of the causal graph: https://www.lesswrong.com/posts/hzuSDMx7pd2uxFc5w/causal-diagrams-and-causal-models ]
+
+Take faces. People are [verifiably very good at recognizing sex from (hair covered, males clean-shaven) photographs of people's faces](/papers/bruce_et_al-sex_discrimination_how_do_we_tell.pdf) (96% accuracy, which is the equivalent of _d_ ≈ 3.5), but we don't have direct introspective access into what _specific_ features our brains are using to do it; we just look, and _somehow_ know. The differences are real, but it's not a matter of any single, simple measurement you could perform with a ruler (like the distance between someone's eyes). Rather, it's a high-dimensional _pattern_ in many measurements you could take with a ruler, no one of which is definitive. [Covering up the nose makes people slower and slightly worse at sexing faces, but people don't do better than chance at guessing sex from photos of noses alone](/papers/roberts-bruce-feature_saliency_in_judging_the_sex_and_familiarity_of_faces.pdf).
+
+Notably, for _images_ of faces, we actually _do_ have transformation technology! (Not "magical", because we know how it works.) AI techniques like [generative adversarial networks](https://arxiv.org/abs/1812.04948) and [autoencoders](https://towardsdatascience.com/generating-images-with-autoencoders-77fd3a8dd368) can learn the structure of the distribution of facial photographs, and use that knowledge to [synthesize faces from scratch](https://thispersondoesnotexist.com/)—or [do things like](https://arxiv.org/abs/1907.10786) sex transformation, as demonstrated by the likes of [FaceApp](https://www.faceapp.com/), the _uniquely best piece of software in the world_.